High-Resolution Gamma Spectrometry of a Plutonium Bearing Waste Drum With High-Energy Reaction-Induced Gamma Rays

In the framework of the radioactive waste drum characterization using neutron coincidence counting, the Nuclear Measurement Laboratory of French Alternative Energies and Atomic Energy Commission (CEA) Cadarache is studying plastic scintillators as an alternative to ideal but costly <sup>3</sup>He gas proportional counters. Plastic scintillators are at least five times cheaper for the same detection efficiency, and in addition, they detect fast neutrons three orders of magnitude more quickly than <sup>3</sup>He detectors. However, they are sensitive to gamma rays with no pulse shape discrimination abilities for large detection volumes, which implies the necessity to identify precisely gamma background sources that may affect the useful signal. This article presents a detailed analysis of the gamma-ray spectrum of a radioactive waste drum containing glove box filters contaminated by plutonium dioxide. Classical gamma emissions following alpha and beta disintegrations are identified, and also those accompanying inelastic scattering (n, <inline-formula> <tex-math notation="LaTeX">$\text{n}^{\prime }$ </tex-math></inline-formula>) or radiative capture (n, <inline-formula> <tex-math notation="LaTeX">$\gamma$ </tex-math></inline-formula>) reactions in the whole waste drum, and (<inline-formula> <tex-math notation="LaTeX">$\alpha $ </tex-math></inline-formula>, n) or (<inline-formula> <tex-math notation="LaTeX">$\alpha $ </tex-math></inline-formula>, p) reactions in the filtration media, which can lead to neutron-gamma coincidences parasitizing useful coincidences from plutonium spontaneous fissions.


I. INTRODUCTION
I N THE context of passive neutron measurements for nuclear material characterization, conventional coincidence collars based on 3 He gas counters permit temporal discrimination of spontaneous fission neutrons and parasitic neutrons from (α, n) reactions. 3 He proportional counters are the gold standard because they show a very high capture cross section for thermal neutrons, leading to high detection efficiency when surrounding detectors with polyethylene moderator, while being practically insensitive to gamma rays. However, the global shortage of 3 He gas [1] has greatly increased the cost in the past 15 years, leading to worldwide initiatives to look for cheaper alternatives with equivalent performances [2]. The Nuclear Measurements Laboratory of French Alternative Energies and Atomic Energy Commission (CEA), Nuclear Energy Division (DEN), Cadarache, is currently studying the use of low-cost polyvinyl toluene (PVT) plastic scintillators as a potential alternative [3], [4] that have the advantage of being sensitive to fast neutrons, thus avoiding thermalizing materials and offering a nanosecond response time. Consequently, very short coincidence windows of a few tens of nanoseconds allow greatly reducing accidental coincidences compared to three orders of magnitude slower 3 He detection systems, mainly due to the time needed for neutron thermalization. However, plastic scintillators are very sensitive to gamma rays. Some of them are emitted in coincidence with (α, n) neutrons often encountered with alpha bearing waste like those contaminated by plutonium and americium oxides, leading to neutron-gamma coincidences that penalize the detection of plutonium spontaneous fission coincidences. Other correlated gamma emissions may also occur in neutron-induced reactions, such as (n, n ) inelastic scattering. Pulse shape discrimination (PSD) liquid and plastic organic scintillators are commonly used to combat gamma sensitivity [5]- [11], but they would not constitute a cost-effective alternative to 3 He counters for large radioactive waste package characterization. Indeed, large detection solid angles and volumes are needed to preserve coincidence counting statistics and therefore, a large number of PSD scintillators should be implemented as their size is limited. Actually, the difference between neutron and gamma-ray pulse tails is not preserved in large PSD scintillators due to multiple light scattering [12], [13]. Finally, a large array of small (maximum 5 in) PSD scintillators would increase the cost to a similar level as 3 He detectors. In addition, PSD information lies in pulse tails and would be damaged by pulse pile-up in case of high count rate, as it is the case with some radioactive waste.
In view to find mitigations to gamma sensitivity of basic PVT plastic scintillators, this article reports the detailed gamma spectrum analysis of a 120-L waste drum containing glove box filters contaminated by plutonium dioxide powder. Classical gamma rays following the radioactive decays of americium and plutonium isotopes are identified, and also those emitted after (α, n) or (α, p) reactions in the filtration media, (n, n ) and (n, γ ) reactions in the whole waste drum.

II. EXPERIMENTAL SETUP
The high purity germanium (HPGe) detector used for the measurement is a planar broad energy germanium (BEGe) from Mirion Technologies (Canberra), equipped with a CP5 Cryo-Pulse. It was in quasi-contact with a 120-L radioactive waste drum containing glove box filters contaminated with plutonium dioxide (PuO 2 ). Table I presents the isotopic composition measured with  multigroup analysis (MGA) codes and Table II presents the mass composition of the filtration media on which the plutonium is deposited. These filters are contained in handles that are attached to a steel frame. Handles are made with polyethylene and thus contain hydrogen, but in an unknown quantity (not mentioned in Table II).
A 1-mm tin plate was placed in front of the detector to clean very low energy gamma rays from the spectrum, especially the very intense 59.5-keV line of 241 Am. The acquisition electronics is a Digital Signal Analyzer LX (DSALX) from Mirion-Canberra, including high-voltage, amplification, pulse digitization and processing, and gamma spectrum storage. Data acquisition and processing are performed with Genie2000 gamma spectrometry software from Mirion-Canberra. The measurement lasted over 53 h, from which 38.7 h of real acquisition time with a dead time of 27%. Registered pulses are coded on 16 384 channels and the resulting spectrum recorded on the [0-6400] keV range is given in Fig. 1.

III. SPECTRUM ANALYSIS
Three main regions of interest are as follows. 1) The well-known region below 800 keV with classical gamma rays following americium and plutonium decays. 2) A region between 800 and 4000 keV including gamma rays emitted by natural radionuclides ( 40 K, thorium and uranium chains), or accompanying (α, x) reactions on the light nuclei of oxygen (plutonium and americium oxides) and of the filtration media, or (n, n ) inelastic scattering reactions, or even radiative capture (n, γ ). This region represents the very heart of this work because reaction gamma rays emitted in correlation with a neutron can lead to real but parasitic neutron-gamma coincidence, which are indistinguishable from useful spontaneous fission coincidences. In the case of (n, γ ) reactions, crosstalk due to Compton scattering can also lead to spurious coincidences. 3) A region between 4000 and 6400 keV with few peaks due to neutron inelastic scattering which can, however, also generate parasitic neutron-gamma coincidence.

A. 0-800-keV Region of Interest
This region, for which a meticulous analysis has already been reported by Sampson [14], is highlighted in Fig. 2. It includes most of the gamma rays following radioactive disintegrations of 241 Am and plutonium isotopes. The two radioactive daughters of 241 Pu, 241 Am, and 237 U, are also present together with plutonium. Some of 237 U gamma rays are clearly visible (164.6, 208.0, 267.5, and 332.36 keV). Fig. 2 also shows the presence of many neptunium and uranium X-rays, 237 Np being produced by alpha decay of 241 Am and beta decay of 237 U. The excited 237 Np daughter nuclei return to ground state by emitting a large number of gamma and X-rays following internal conversion. 234 U, 235 U, and 236 U isotopes are produced from the alpha decay of 238 Pu, 239 Pu, and 240 Pu isotopes, leading also to many gamma and X-rays. 241 Pu (together with its 241 Am and 237 U daughters) is the major contributor to the low-energy part of the spectrum (up to the 335.38-keV peak), followed by 239 Pu and 238 Pu. With a long half-life of 373 × 10 3 years and small intensity gamma or X-rays, 242 Pu has no visible signature on the spectrum.

B. 800-4000-keV Region of Interest
Figs. 3 and 4 show the spectrum in the 800-1900-and 1900-4000-keV ranges, respectively. Tables III and IV give the intensity and the emitter of each identified gamma ray. Fig. 3 also shows many signatures from the 232 Th natural decay chain, notably from 228 Ac, 212 Bi, and 208 Tl. The 228 Ac radionuclide shows many peaks at 911, 965, 969, 1495, 1588, and 1631 keV. The gamma rays of 212 Bi (1079 and 1513 keV) and 208 Tl (861-keV line, 2614 keV plus its single escape (SE) and double escape (DE) peaks) are also clearly detected. From the 238 U natural decay chain, seven signatures of 214 Bi are also visible at 1120, 1408, 1509, 1661, 1730, 1764, and 2204 keV, in addition to the 609-keV gamma ray clearly visible on Fig. 2. Finally, the classical 1461-keV gamma ray of natural 40 K is also present.
In addition to the above classical lines, the 800-4000-keV region also shows less common gamma rays originating from alpha and neutron-induced reactions, which are of high importance for our neutron-gamma coincidence measurements  with plastic scintillators. Excited states of reaction product nuclides have been analyzed to propose potential reactions that can lead to the observed gamma rays. Energy levels and gamma-ray transitions are taken from the Evaluated Nuclear Structure Data File (ENSDF) [15]. Cross sections of (α, x) reactions were calculated with EMPIRE-II v.2.18 code [16] developed by the International Atomic Energy Agency (IAEA), considering 5.5-MeV alpha particles. Cross sections  [17], where experimental data have been reported for 14.5-MeV neutrons. Cross sections for thermal neutron radiative capture and corresponding gamma-ray energies are taken from [18].
The following list, in the order of increasing atomic number of the target nuclei or excited reaction daughters (e.g., 13 C, 22 Na, and 26 Mg), reports capture gamma rays and gamma transitions of (α, x) and (n, n ) reaction products that possibly correspond to gamma rays observed in the spectrum.
1) The well-known 2223-keV gamma ray from thermal neutron radiative capture by 1 H [18] is clearly visible in the spectrum, hydrogen being present in plastic handles in which the glove box filters are conditioned. 2) Two intense gamma rays corresponding to transitions from the two and three excited levels of 13 C (produced by (α, p) reaction on 10 B as boron are present in the filtering media) to the ground state and are visible in Fig. 4 [19]. These two lines are 3684 keV [3684.507 . Large Doppler energy broadening is observed for the 3684-keV peak because the short lifetime of the excited state (half-life of 1.10 fs) allows gamma ray deexcitation while the residual 13 C nucleus is still slowing down. The Stopping and Range of Ions in Matter (SRIM) [20] calculation indeed shows that the slowing down of a 13 C nucleus is typically hundreds of femtoseconds. The shorter the excited state lifetime, the larger the broadening effect, as can be seen with the 3853-keV gamma ray, for which the de-excitation of the 3853.807-keV level (half-life of 8.60 ps) induces a smaller broadening. The 3089-keV gamma ray corresponding to the transition from the first excited state of 13 C to the ground state was also expected but is not identifiable in Fig. 4, probably due to an expected large Doppler broadening (1.07-fs lifetime of the first excited level).
3) The transition between the first excited state and the ground state of 17 O [870.73 (1/2 + ) → 0 (5/2 + )] leads to an 871-keV gamma ray [21] visible in the spectrum. Due to the long lifetime of this first excited level (179 ps), no Doppler effect should be observed but the base of the peak is clearly broadened. This might be due to another unidentified broadened transition close to 870 keV. This gamma ray might be produced by inelastic scattering of alpha or neutron particles on 17 O nuclei. However, as 17 O is the least abundant isotope of oxygen (0.038%), the (n, n ) production mode is negligible because only elements with a minimum weight proportion of a few percent can lead to detectable fast neutron-induced gamma rays [22]. Conversely, the (α, α ) reaction on 17 19 F excited levels should also lead to low-energy gamma rays, but in this energy region they are masked by the  Fig. 4. Table V gives the six most intense deexcitation gamma rays of this nucleus. The 351-keV gamma ray is not visible because it is masked by americium and plutonium gamma emissions. The 1120and 1395-keV gamma rays may also be present, but the 214 Bi gamma ray at 1120 keV, on the one hand, and the broadened peak at 1400 keV from 22 Na (as explained below), on the other hand, prevent from identifying them. Oxygen is very abundant in the waste and 21 Ne is probably produced by (α, n) reactions on 18 (Fig. 3) and 2083 keV (Fig. 4) gamma rays [25]. Both peaks show Doppler broadening, which is consistent with the short lifetime of the 3357.2-keV level (225 fs) for the 2082-keV peak, but not with the longer lifetime (3600 fs) of 1274.537-keV level for the 1274-keV peak. Doppler broadening is probably due to another nuclear reaction, discussed later in this article, producing the 29 Si first excited level with a short lifetime (291 fs) and leading to a 1273-keV gamma ray. 7) Among the first height most intense (100% of relative intensity) de-excitation gamma rays given in Table VI for 22 Na [25], four are observed on the spectrum: 891, 1400, 1528, and 1554 keV. The 583-keV gamma ray is interfered by the 583-keV gamma ray of 208 Tl, and the 74 keV is masked by the intense americium and plutonium signal. The 1280 keV is probably drowned in the very intense broadened peak at 1274 keV, which is also contributed by the beta decay of 22 Na. Concerning the 1369-keV peak, because of the very short half-life of the 1951.8-keV level (8 fs), an important broadening is expected and might be responsible for the fact that this gamma ray is not clearly identifiable in Fig. 3. However, a broad bump is clearly present, probably merged with the broad peak of 19 F at 1357 keV mentioned earlier. The 22 Na excited state probably comes from the (α, n) reaction on 19 F. The 511-keV annihilation shown in Fig. 2 is a signature of the beta decay of generated 22 Na nuclei. 8) The 975-keV peak visible in Fig. 3 is assigned to the [974.756 (3/2 + ) → 0 (5/2 + )] transition of 25 Mg [26] produced by (n, α) reaction on 28 Si. Indeed, silicon is a major constituent of the waste (see Table II) and 28 Si presenting a 92 % abundance. On the other hand, magnesium is present in the filter waste composition but in limited weight fraction (0.71%, see Table II), 25 Mg has only a 10% isotopic abundance, and neutron inelastic scattering could only be a minor production mode. In case of sufficient neutron thermalization in the waste, radiative capture by 24 Mg (79% isotopic abundance in magnesium) might also be a way to produce excited 25 Mg nuclei, but with a very small cross section of 53.7 mb [18] making this production negligible too. Nuclear data for 25 Mg energy levels and gamma transitions show that the 585 keV (100% of relative intensity) and 1611 keV (100% of relative intensity) gamma rays should also be present in the  [26]. However, the first one is in interference with the 583-keV peak of 208 Tl, and the second one comes from an excited state with a very short half-life [1611.768 keV (7/2 + , 17 fs)] potentially leading to large Doppler broadening detrimental to its detection. 9) Several gamma signatures from 26 Mg are observed.
The main gamma transitions of this isotope are given in Table VII [27]. Excited levels of 26 Mg can be produced by (α, p) reactions on 23 Na nuclei present in the waste. On the contrary, neutron inelastic scattering on 26 Mg is probably negligible as this isotope abundance is only 11% of magnesium, this last representing a small mass fraction (0.71%) of the waste composition. Seven gamma rays given in Table VII are visible on the spectrum. Most of these are due to transitions toward the 1808.74-keV first excited level, which explains the high intensity of the 1809 keV that is also largely Doppler broadened due to its short half-life (476 fs). These cascades of gamma transitions temporally correlated with the 1809-keV gamma ray will again lead to parasitic coincidence with respect to spontaneous fission coincidences in plastic scintillators. The 1003-keV gamma ray interferes with the well-known 234m Pa gamma ray at 1001 keV ( 238 U chain), but it seems to be present on the right side of this peak. The 1130-and 2938-keV gamma rays, coming from the same excited level at 2938.33 keV with a short half-life (141 fs), present a large energy broadening by Doppler effect. A lower relative intensity and a weaker detection efficiency for the 2938.15-keV gamma ray, compared to 1129.61-keV gamma ray, lead to the fact that only the broadening around 2938 keV is identifiable on Fig. 4. The 1780-keV transition is also visible in Fig. 3, but other reactions leading to excited 28 Si can also contribute at 1779 keV, as discussed in the following. Excited levels that can be populated, such as 2938.33-, 4318.89-, or 4350.09-keV levels, may induce a correlation between several gamma rays. For example, an (α, p) reaction leading to the 4350.09-keV excited state of 26 Mg can induce a correlated gamma cascade with the 1412-, 1130-, and 1809-keV gamma rays, and subsequently to coincidence possibly disturbing spontaneous fission coincidence counting. On the other hand, the 1897-, 2524-, and 2541-keV gamma rays reported in Table VII are not identified on the spectrum (see Fig. 4). 10) The 844-and 1014-keV gamma rays (both with a 100% relative intensity) visible in Fig. 3 are characteristic of 27 Al [28]. The 2212-keV transition (100% relative intensity) expected for this excited nuclei is not visible because of the proximity to the broadened 2235-keV peak due to 27 Al(α, p) 30 Si, as detailed further. The 27 Al excited nuclei come from (α, p) reactions on 24 Mg, which has a 79% isotopic abundance in natural magnesium. Alpha radiative capture on 23 Na or alpha inelastic scatterings are discarded because of very low cross sections (respectively ∼10 −2 and 1 mb) compared to (α, p) cross section (∼20 mb) given in Table III. Furthermore, excited states responsible for these observed gamma rays cannot be populated by inelastic scattering of 5.5-MeV alpha particles [28]. 11) A single transition of 1779 keV from the first excited state of 28 Si (1779.030 (2+), 475 fs of half-life) to the ground state [29] is identified in Fig. 3. The excited 28 Si nuclei is produced by neutron inelastic scattering on 28 Si (92.2% isotopic abundance in silicon) or (α, p) reaction on 25 Mg. As mentioned earlier, the 1779-keV line is in interference with the 1780-keV gamma ray due to 23 Na(α, p) 26 Mg reaction. 12) The 1273-keV gamma ray from the transition between the first excited state [1273.387 (3/2+)] and the ground state (1/2+) of 29 Si [30] is present in Fig. 3. Radiative capture on 28 Si (92.2% isotopic abundance) and inelastic scattering on 29 Si (4.7% abundance) can lead to excited 29 Si, as well as 26 Mg(α, n) and 29 Si(α, α ) reactions. The 1273-keV peak shows significant Doppler broadening due to the short half-life (291 fs) of the 1273.387-keV excited level. It is worth reminding that this peak is very close to the already quoted 1274-keV gamma ray of 22 Ne. As mentioned earlier, this last is not Doppler broadened, in principle, due to the long half-life of its original 22 Ne excited level, and therefore broadening only comes from the 29 Si level. 13) Two gamma rays of 30 Si are identified in Fig. 4 at 2235 keV, from the first excited level (215 fs) to the ground state, and 3498-keV gamma ray from the second excited level (58 fs) to the ground state [31] (see Table VIII). Both are logically Doppler broadened due to the short lifetimes of their original excited levels, especially the 3498-keV gamma ray for which the full energy peak is not even visible. The 1263-keV gamma transition is almost completely drowned in the intense and broadened 1274-keV peak, but it might be guessed on the left tail. The 2595-keV transition originating from the 4830.84-keV level (83 fs) is expected to be very broadened and seems to be present on the left tail of the intense 2614.533-keV peak of 208 Tl. Other transitions presented in Table VIII are not observable on the spectrum, most of them being expected with large Doppler  broadening. The a priori unbroadened 1552-keV peak might be merged with the 1554-keV peak due to 19 F(α, n) 22 Na, as reported earlier. The excited 30 Si levels can be produced by inelastic scattering of neutrons or alpha particles, silicon being a major chemical element in the waste with a 30 Si isotopic abundance of 3%. It can also be produced by (α, p) reaction on 27 Al. 14) Four gamma rays corresponding to de-excitation of 36 Cl nuclei produced by neutron radiative capture on 35 Cl (76% abundance in chlorine and large cross section of 43.6 b [18]) are identified in the spectrum, as reported in Table IX [32]. The 788-and 517-keV gamma rays are probably masked by the intense americium and plutonium signal. The 1601 keV (very close to the intense 1592-keV peak of 208 Tl), 2518 keV (maybe masked by the 2510.01-keV Doppler broadened peak due to 23 Na(α, p) 26 Mg), and 2676 keV (large Doppler broadening expected due to the short lifetime of the excited level, 21 fs) are also not visible. 15) The 847-keV transition between first excited level and ground state of 56 Fe is identified in the spectrum, as well as the 1238-keV transition between its two and one excited levels [33]. The high atomic number of iron does not allow alpha particles to get through the Coulomb barrier, so neutron inelastic scattering on 56 Fe is the main reaction responsible for these two gamma rays. In addition to the neutron and gamma correlation in the (n, n ) reaction, the cascade between the 1238-and 847-keV gamma rays can lead to the emission of up to three time-correlated particles, potentially leading to high order parasitical coincidences with respect to plutonium spontaneous fission coincidences.

C. 4000-6400-keV Region of Interest
The high-energy region of the spectrum in Fig. 5 shows neutron-induced radiative capture and inelastic scattering gamma rays on carbon and oxygen nuclei (see Table X) that are present inside the waste drum. The 4946.31-keV radiative capture gamma ray on 12 C nuclei is weak but present, as well as the 6128.63-keV inelastic scattering gamma ray on 16 O accompanied by its SE and DE peaks. The 16 O(n, n ) reaction may also lead to 6917-and 7117-keV gamma rays [17] but they are beyond the 6400-keV end-energy of the spectrum. After a discussion with the detector manufacturer, the artifact peak near 5500 keV is probably due to a connection issue of the "energy" cable between the preamplifier and the digital electronics. This peak was also present in the background noise spectrum. It is important to note that some peaks of this region remain to be identified (e.g., near 4700 and 5500 keV).

IV. CONCLUSION
The detailed gamma-ray spectrometry analysis of a 120-L waste drum filled with plutonium bearing glove box filters allowed us to identify, besides natural background gamma and X-rays ( 40 K, U, and Th chains) and well-known gamma and X-rays due to plutonium and americium radioactive decays, a number of less common gamma rays. These remaining rays are due to nuclear (α, x) reactions induced by the high alpha activity of Pu and Am, as well as by neutrons from spontaneous fissions and (α, n) reactions. The analysis of the residual nuclei excited levels produced in alphaand neutron-induced reactions allowed us to identify several unusual gamma rays and the potential spurious coincidences that they could produce in a neutron multiplicity measurement system with plastic scintillators. The observed lines, as well as the Maxwellian distribution of spontaneous fission neutrons, will be used in further Monte Carlo N-Particles (Transport Code) (MCNP) simulations to compare pulse-height spectra of energies deposited by neutrons and gamma rays in plastic scintillators. It will allow us to estimate the interest of placing low-and high-energy rejection thresholds to limit the influence of parasitic coincidence involving gamma rays.